Statistical multiplexing of self-similar video streams: simulationstudy and performance results

摘要

Achieving statistical gains when multiplexing video streams, as inna video-on-demand (VOD) scenario, is difficult because of the stringentnQOS demands and the self-similar nature of the traffic. This papernexplores, through empirical simulation, the QOS, network utilization,nand statistical characteristics of the aggregate traffic resulting fromnmultiple independent MPEG video streams. In addition, the simulationnresults are compared against several recently-derived theoreticalnresults for self-similar network traffic. Three main results are evidentnfrom our experiments. First, moderate statistical multiplexing gain cannbe achieved when multiplexing multiple self-similar streams. Second,nvideo multiplexing is extremely sensitive to traffic phasing effects andnto heavy-tailed frame size distributions. Finally, the theoreticalnapproach considered (Norros (see IEEE Journal on Selected Areas innCommunications, vol.13, no.6, p.953-62, 1995) effective bandwidthnformulation) appears promising but requires fine-tuning to be practicalnfor call admission and network dimensioning